Cost analysis of a school-based comprehensive malaria program in primary schools in Sikasso region, Mali

The vast majority of malaria morbidity and mortality occurs in sub-Saharan Africa, where infection with Plasmodium falciparum remains a considerable obstacle to both health and economic development. Traditionally, efforts have targeted maternal and child morbidity and mortality driven by the persistent malaria burden during pregnancy and early childhood. However, an increasing body of evidence points to broader social and economic impacts of the disease [1‐7]. In particular, there is increased recognition of the value of including of school-aged children in these efforts, due to the high parasite rates in this population and the increasing body of evidence demonstrating developmental consequences of infection throughout childhood and adolescence [8, 9].

Malaria has been associated with reductions in sustained attention, cognition, and school achievement in various settings, including Mali, where this study is based. Analysis of data from demographic and health surveys found that a higher prevalence of malaria in a community was associated with higher primary repetition rates [3]. In addition, a longitudinal survey in an area with high and stable malaria transmission area has shown that Plasmodium falciparum infection affects educational achievement and cognitive performance [4]. Similar results have been observed elsewhere [5‐7].

School-based delivery of interventions represents a potential strategy to address this burden. A cluster-randomised trial in western Kenya demonstrated schools as a promising and cost-effective channel through which to deliver malaria treatment to schoolchildren and promote improvements in attention and school performance [6, 10]. There are few data on the cost or cost-effectiveness of school-based interventions. A systematic review in 2011 identified 48 studies that evaluated the cost-effectiveness of malaria interventions [11], of which only two [10, 12], were conducted among school-aged populations; as well as a cost analysis of a more recent school-based intervention in Kenya [13]. Better understanding the financial and economic costs of school-based interventions will help to inform ongoing questions of whether malaria control programmes can be channelled through schools, particularly in light of the availability of new preventive strategies such as seasonal malaria chemoprevention [14, 15], and whether integrated programmes rather than single interventions should be implemented [10, 13, 16].

The main objective of this study is to describe the costs of a comprehensive school-based malaria intervention strategy carried out in an area of highly seasonal malaria transmission in Mali, following an ingredients approach. The intervention included three components: participatory malaria education; distribution of long-lasting insecticide-treated nets (LLIN); and a 3-day course of treatment with artemisinin-combination therapy administered at the end of the transmission season to all schoolchildren (intermittent parasite clearance in schools, or IPCs). Key cost drivers are examined to inform a discussion of factors to consider in implementing or scaling up the programme at the national level.

Various factors must be considered in interpreting the costs of the school-based intervention described here. The first is the difference between a comprehensive intervention, as presented here, and other cost models for malaria control available in the literature which tend to focus on education, LLIN distribution, or treatment separately. Whereas in combined programmes cost savings and economies of scope can be generated by a joint delivery, the effectiveness of this programme in this area of seasonal malaria depends on the correct sequential timing of the activities: LLINs and malaria education are delivered ahead of the onset of the annual rains to protect schoolchildren from new infections during the period of highest risk when schools are closed, whereas IPCs is given several months later at the end of the transmission season to clear malaria parasites and reduce malaria-related anaemia. This results in a conservative bias of our estimate, although we suggest that combining some activities would plausibly result lower costs. For example, training costs and opportunity costs would be reduced if the training session for teachers occurred only once, and the combined length decreased by a full day. The second point to consider is how these costs might differ under conditions of programme adoption or replication by government agencies, both within this context and in other comparable settings. Costs were split between start up costs, equal to 38% of the total, and operational costs (post start-up) that represent 62% of the intervention. The breakdown suggests that if the programme were to be repeated in the area with the same teachers, there would be cost saving in planning and training activities in subsequent years. The impact on costs attributable to implementation by an INGO, which supported all planning, management and logistical expertise, must be considered in comparison with other implementation bodies. The level of remuneration and the number of staff deployed for the intervention might not be affordable by other actors with fewer resources. Similarly, one of the factors that made training activities expensive was the payment of per diem and transport allowance to the participants. High level of training attendance is crucial for the implementation of the intervention and financial incentives are important in context of limited resources. However, one should consider if the rate set for the allowances was appropriate compared to teachers’ salary. In this study, teachers’ daily salary was 9 USD, while daily per diem was 11.75 USD. Finally, differences within the country are important to consider should the program be rolled out in other regions. Heterogeneities in factors such as accessibility by road, distance between schools, and region size will impact transport and logistical costs. In terms of human resources, excluding Kidal region, the ratio between teachers and children is fairly consistent throughout the country, with a national average of 1:50 [26]. This is larger than the ratio in the study (1:36), highlighting another important area of potential cost-savings given that teachers’ salaries accounted for 15% of total costs reported here. These are among the details warranting further consideration should the government of Mali choose to integrate this intervention into a national strategy. Finally, given that the policy of universal LLIN distribution is now common practice in the Sahel region, removal of the LLIN component from the school program represents a likely scenario for governments in the region. Thus, when considering the replication of a similar school-based intervention in a context with this policy, the LLIN component should be re-evaluated. If removed, our data show that cost per child would decrease to 4.89 USD. This effectively halves the costs of intervention; and is a potentially important finding for policymakers considering complementary strategies to address the burden of malaria among school-aged children.

Additional factors need also be considered in extrapolating these results to other settings. The external validity of these findings may be dependent on factors affecting both the costs and likely effectiveness of this approach. These include, but are not limited to, the dispersion of schools throughout a region, school enrolment rates and regularity of attendance, urban versus rural settings, the seasonality of malaria transmission, and the relative burden of malaria, which was high in this community.

The delivery strategy adopted has a number of potential merits. A major advantage of school-based interventions is availability of personnel, which would be difficult to achieve for programmes requiring health staff in remote settings. The intervention also achieved very high coverage, amongst children enrolled in school. This represents a key strength of the school-based approach in rural, low-resource settings. However, it is important to note the potential negative impacts of relying on school infrastructure to deliver health programmes. A more detailed understanding of the opportunity costs of taking teachers’ and schoolchildrens’ time and re-allocating it to health activities, particularly in settings whose education system already produces poor results, is needed.

Finally, we consider how robust these results are. Sensitivity analyses showed that variation in LLIN price, reduction in salaries for those activities performed by STC personnel, and levels of wastage could all have a substantial impact on cost. All of these aspects can be influenced by the capacity of the actor managing the intervention. A reduced LLIN unit price might be obtained by linking the procurement for this intervention to other national campaigns, and a national authority might be in a better position to do so. In terms of wastage, the intervention showed very low rates that could be attributed to good organization and vigilant management.

Improving child health remains a policy concern in Sub-Saharan Africa and an important objective for the global health community, demanding both innovative and feasible approaches. This study has provided a cost analysis of a comprehensive school-based malaria intervention, in a rural setting with highly seasonal transmission. As in other school-based interventions, teachers represented the key human resources in the actual implementation of the intervention activities, and were able to achieve a high level of coverage. Human resources were estimated as the key cost drivers. Training was the activity that absorbed the largest portion of resources in the intervention as delivered; followed by cost of purchasing LLINs. From a provider perspective, evaluating the total expenditure of 10.38 USD (actual intervention) or 4.85 USD (universal LLIN scenario) per child as an affordable strategy to combat malaria would have to consider the effectiveness, including the health gain and the effects on school outcome of the intervention. Clarke et al. [17] found strong evidence for an impact of the intervention on malaria infection, anaemia, and haemoglobin concentration, as well as significant improvements in cognitive function, and this effect was found to endure throughout the low-transmission period. Taken together with the epidemiological findings, the current cost analysis provides impetus for the consideration of comprehensive school-based malaria control programs to target the burden of malaria in school-aged children in the Sahel region.